Post by thread:[EE] Good general purpose voltage reference

{Quote hidden}>-----Original Message----- >From: spam_OUTpiclist-bouncesTakeThisOuTmit.edu [.....piclist-bouncesKILLspam@spam@mit.edu] >On Behalf Of Glenn Jones >Sent: 17 November 2004 09:29 >To: piclistKILLspammit.edu >Subject: [EE] Good general purpose voltage reference > > >Hello, >I am looking for suggestions for a good reliable general >purpose voltage reference chip. The voltage I need eventually >is 2.5 V, but I could divide down from 4.096 or the like. The >system has a 5V supply. I don't need anything too accurate, >+-.01V would be good. If possible, I'd like something that has >been around a while so that I can hope that it will be >available for 10-20 years, or at least a pin compatible >replacement. Is there a standard voltage reference pinout? >Temperature stability is also important, but I'm not sure >exactly how much so. Is 100 ppm/deg C unreasonable? Thanks for >any advice, Glenn _______________________________________________ Virtualy all references meet or better 100ppm. If you want to pay for it you can get better than 2ppm references. However, I guess we'll need some more requirements before suggesting parts. Do you want through hole or surface mount parts? What current capability do you need? (The PIC's A/D can take quite large currents from the reference). I cost a big issue? Regards Mike ======================================================================= This e-mail is intended for the person it is addressed to only. The information contained in it may be confidential and/or protected by law. If you are not the intended recipient of this message, you must not make any use of this information, or copy or show it to any person. Please contact us immediately to tell us that you have received this e-mail, and return the original to us. Any use, forwarding, printing or copying of this message is strictly prohibited. No part of this message can be considered a request for goods or services. ======================================================================= ____________________________________________

National Semiconductor LM385-2.5 ??? Regards Attila - SM4RAN ______________________________________________________ Gratis visitkort - klicka här ! Allt inom e-handel! http://www.torget.se -----Ursprungligt meddelande----- Från: Glenn Jones <.....lynxlabKILLspam.....gmail.com> Till: EraseMEpiclistspam_OUTTakeThisOuTmit.edu <piclistspam_OUTmit.edu> Datum: den 17 november 2004 10:46 Ämne: [EE] Good general purpose voltage reference {Quote hidden}>Hello, >I am looking for suggestions for a good reliable general purpose >voltage reference chip. The voltage I need eventually is 2.5 V, but I >could divide down from 4.096 or the like. The system has a 5V supply. >I don't need anything too accurate, +-.01V would be good. If possible, >I'd like something that has been around a while so that I can hope >that it will be available for 10-20 years, or at least a pin >compatible replacement. Is there a standard voltage reference pinout? >Temperature stability is also important, but I'm not sure exactly how >much so. Is 100 ppm/deg C unreasonable? >Thanks for any advice, >Glenn >____________________________________________

{Quote hidden}>> >-----Original Message----- >> >From: KILLspampiclist-bouncesKILLspammit.edu [RemoveMEpiclist-bouncesTakeThisOuTmit.edu] On >> >Behalf Of Glenn Jones >> >Sent: 17 November 2004 09:29 >> >To: spamBeGonepiclistspamBeGonemit.edu >> >Subject: [EE] Good general purpose voltage reference >> > >> > >> >Hello, >> >I am looking for suggestions for a good reliable general purpose >> >voltage reference chip. The voltage I need eventually is >2.5 V, but I >> >could divide down from 4.096 or the like. The system has a >5V supply. >> >I don't need anything too accurate, >> >+-.01V would be good. If possible, I'd like something that has >> >been around a while so that I can hope that it will be >available for >> >10-20 years, or at least a pin compatible replacement. Is there a >> >standard voltage reference pinout? Temperature stability is also >> >important, but I'm not sure exactly how much so. Is 100 ppm/deg C >> >unreasonable? Thanks for any advice, Glenn >> >_______________________________________________ >On Wed, 17 Nov 2004 09:55:29 -0000, Michael Rigby-Jones wrote: >> >> Virtualy all references meet or better 100ppm. If you want >to pay for >> it you can get better than 2ppm references. However, I guess we'll >> need some more requirements before suggesting parts. Do you want >> through hole or surface mount parts? What current capability do you >> need? (The PIC's A/D can take quite large currents from the >> reference). I cost a big issue? >-----Original Message----- >From: TakeThisOuTpiclist-bouncesEraseMEspam_OUTmit.edu [RemoveMEpiclist-bouncesTakeThisOuTmit.edu] >On Behalf Of Glenn Jones >Sent: 17 November 2004 10:47 > >Hello, >Thank you for the reply. >I have no preference for through hole vs. surface mount. I do >not expect any large current capability from the reference, I >intend to buffer it with an op-amp, but if there are >references out there that can supply 10 - 50 mA, that'd be >great. Thanks, Glenn Firstly be carefull when buffering, the input offset of an op-amp can easily have a worse tempco than your voltage reference. Have a look at the Linear Technology LT1461 (www.linear.com/pc/productDetail.do?navId=H0,C1,C1154,C1002,C1243,P20 76 ). This is availble with 2.5volt output and can source up to 50mA, tempco 3ppm. Not cheap though, $1.95 in single quantities for the cheapest variant. Analog Devices make the RE192 ( http://www.analog-devices.com/en/prod/0,,769_838_REF192%2C00.html ). Tempco 5ppm, output current 30mA, cost $1.24/1000 Unfortunately the high output current, low tempco and low cost reference doesn't seem to exist IME, but I'd be happy to learn otherwise! Mike ======================================================================= This e-mail is intended for the person it is addressed to only. The information contained in it may be confidential and/or protected by law. If you are not the intended recipient of this message, you must not make any use of this information, or copy or show it to any person. Please contact us immediately to tell us that you have received this e-mail, and return the original to us. Any use, forwarding, printing or copying of this message is strictly prohibited. No part of this message can be considered a request for goods or services. ======================================================================= ____________________________________________

Glenn Jones wrote: > I am looking for suggestions for a good reliable general purpose > voltage reference chip. The voltage I need eventually is 2.5 V, but I > could divide down from 4.096 or the like. The system has a 5V supply. > I don't need anything too accurate, +-.01V would be good. That's .4%, which is actually quite accurate. There are lots of choices at 1% and 2%, but less than .5% will limit the field considerably. > Temperature stability is also important, but I'm not sure exactly how > much so. Is 100 ppm/deg C unreasonable? No, that's quite a bit. Note however that a 40degC variation blows your entire error budget. For example, if you can only guarantee the temperature stays within a 20degC range, then you need .2% initial accuracy. That's going to be expensive. ***************************************************************** Embed Inc, embedded system specialists in Littleton Massachusetts (978) 742-9014, http://www.embedinc.com ____________________________________________

Glenn Jones wrote: > I have no preference for through hole vs. surface mount. I do not > expect any large current capability from the reference, I intend to > buffer it with an op-amp, but if there are references out there that > can supply 10 - 50 mA, that'd be great. First, 50mA is "large" current for a voltage reference. Second, buffering with an opamp is going to introduce additional errors. You wanted the result to be 2.5V within 10mV. The opamp will eat up a millivolt or two due to its offset voltage. You should step back and explain the overall problem. You seem to have jumped into the middle specifying something without understanding the tradeoffs and what's out there. It sure sounds like this problem needs to be solved with a more global look at the circuit. ***************************************************************** Embed Inc, embedded system specialists in Littleton Massachusetts (978) 742-9014, http://www.embedinc.com ____________________________________________

At 01:29 AM 11/17/2004 -0800, you wrote: >Hello, >I am looking for suggestions for a good reliable general purpose >voltage reference chip. The voltage I need eventually is 2.5 V, but I >could divide down from 4.096 or the like. The system has a 5V supply. >I don't need anything too accurate, +-.01V would be good. If possible, >I'd like something that has been around a while so that I can hope >that it will be available for 10-20 years, or at least a pin >compatible replacement. Is there a standard voltage reference pinout? >Temperature stability is also important, but I'm not sure exactly how >much so. Is 100 ppm/deg C unreasonable? >Thanks for any advice, >Glenn Hi, Glenn:- Here are some general suggestions, but there are lots of other possibilities depending on the exact requirements. Note that if you use another voltage and divide it down or buffer it you'll need to consider resistor errors and drift or offset voltage etc. in the error budget. Buffering for the PIC ADC reference requires a very low impedance source; probably a bit more than a simple op-amp output, like a couple of resistors and a cap to get a stable low-Z source. LM4120AIM5-2.5 +/-0.2% initial accuracy +/-50ppm/K series voltage reference. $1.00 1K / 1.62/100 160uA typical draw (has 2uA shutdown) LM4040BIM3-2.5 +/-0.2% initial accuracy +/- 100ppm/K shunt voltage reference $0.62 1K / 0.87/100 (needs about 100uA to operate) LM431CCZ +0.5%/-0.6% initial accuracy +/-50ppm/K shunt voltage reference $0.33 1K/ 0.51/100 (needs about 1mA to operate) The LM431 is made in enormous volume by many suppliers, but most versions are only +/-1% accurate, and it needs about 1mA to regulate. The lower accuracy versions get down to the 10-cent range in medium volume. Best regards, Spehro Pefhany --"it's the network..." "The Journey is the reward" speffEraseME.....interlog.com Info for manufacturers: http://www.trexon.com Embedded software/hardware/analog Info for designers: http://www.speff.com ____________________________________________

Spehro Pefhany wrote: > Here are some general suggestions, but there are lots of other > possibilities depending on the exact requirements. Note that if you > use another voltage and divide it down or buffer it you'll need to > consider resistor errors and drift or offset voltage etc. in the > error budget. Buffering for the PIC ADC reference requires a very > low impedance source; probably a bit more than a simple op-amp output, > like a couple of resistors and a cap to get a stable low-Z source. Or you can feed the 2.5V reference into the A/D as another input, run the A/D reference off the Vdd supply, and do the correction in firmware. This is what I did in the ProProg. The 2.5V reference from a LM385 goes directly to an A/D input. The reference voltage is still read with about 1 part in 500 resolution, which is well beyond its accuracy. Multiplying it up to a higher voltage would have introduced much more error than it would have given back in terms of A/D resolution. Correcting the other readings to the measured reference is just a little math in the firmware, which costs nothing in parts and introduces virtually no additional error. ***************************************************************** Embed Inc, embedded system specialists in Littleton Massachusetts (978) 742-9014, http://www.embedinc.com ____________________________________________

On Wed, 2004-11-17 at 10:29 -0500, Olin Lathrop wrote: > Or you can feed the 2.5V reference into the A/D as another input, run the > A/D reference off the Vdd supply, and do the correction in firmware. This > is what I did in the ProProg. The 2.5V reference from a LM385 goes directly > to an A/D input. The reference voltage is still read with about 1 part in > 500 resolution, which is well beyond its accuracy. Multiplying it up to a > higher voltage would have introduced much more error than it would have > given back in terms of A/D resolution. Correcting the other readings to the > measured reference is just a little math in the firmware, which costs > nothing in parts and introduces virtually no additional error. Wow, that's a really cool idea! Thanks Olin, I'll store that one for later use. TTYL ____________________________________________

At 10:29 AM 11/17/2004 -0500, you wrote: >Spehro Pefhany wrote: > > Here are some general suggestions, but there are lots of other > > possibilities depending on the exact requirements. Note that if you > > use another voltage and divide it down or buffer it you'll need to > > consider resistor errors and drift or offset voltage etc. in the > > error budget. Buffering for the PIC ADC reference requires a very > > low impedance source; probably a bit more than a simple op-amp output, > > like a couple of resistors and a cap to get a stable low-Z source. > >Or you can feed the 2.5V reference into the A/D as another input, run the >A/D reference off the Vdd supply, and do the correction in firmware. This >is what I did in the ProProg. The 2.5V reference from a LM385 goes directly >to an A/D input. The reference voltage is still read with about 1 part in >500 resolution, which is well beyond its accuracy. Multiplying it up to a >higher voltage would have introduced much more error than it would have >given back in terms of A/D resolution. Correcting the other readings to the >measured reference is just a little math in the firmware, which costs >nothing in parts and introduces virtually no additional error. That sounds like a good solution for that problem. Depending on what you're doing, the resolution and quantization noise may be very important. Worst case, without dither and digital low-pass filtering, you could have the reference bobbling up and down a count and the input bobbling up and down a count, which would represent a +/-0.8% variation. Not acceptable if you're trying to display engineering units, such as 800°F or 1000 PSI. It would also degrade the derivative term in PID control (requiring more filtering). There are also possible issues with Vdd variations if there are heavy loads from the Vdd bus (a device I'm developing at the moment has a load that varies from 0 to 2A with fairly sharp edges. It would be possible to run the micro on a separately regulated 3.3V supply to reduce short-term variations in Vdd for the micro, but that results in additional complexity with the mixed voltages on the board.) Hopefully Microchip will soon figure out (like several other companies have been doing for years) how to put proper delta-sigma ADCs (and dare I ask for a PGA option?) into inexpensive PICs so we won't have to deal with crummy 10 bit resolution for much longer. With a 16-24 bit ADC option, then we can really forget about resolution issues and concentrate on accuracy. Best regards, Spehro Pefhany --"it's the network..." "The Journey is the reward" EraseMEspeffinterlog.com Info for manufacturers: http://www.trexon.com Embedded software/hardware/analog Info for designers: http://www.speff.com ___________________________________________

Excellent! -----Original Message----- From: RemoveMEpiclist-bouncesEraseMEEraseMEmit.edu [RemoveMEpiclist-bouncesspam_OUTKILLspammit.edu] On Behalf Of Olin Lathrop Sent: Wednesday, November 17, 2004 8:30 AM To: Microcontroller discussion list - Public. Subject: Re: [EE] Good general purpose voltage reference Spehro Pefhany wrote: > Here are some general suggestions, but there are lots of other > possibilities depending on the exact requirements. Note that if you > use another voltage and divide it down or buffer it you'll need to > consider resistor errors and drift or offset voltage etc. in the > error budget. Buffering for the PIC ADC reference requires a very > low impedance source; probably a bit more than a simple op-amp output, > like a couple of resistors and a cap to get a stable low-Z source. Or you can feed the 2.5V reference into the A/D as another input, run the A/D reference off the Vdd supply, and do the correction in firmware. This is what I did in the ProProg. The 2.5V reference from a LM385 goes directly to an A/D input. The reference voltage is still read with about 1 part in 500 resolution, which is well beyond its accuracy. Multiplying it up to a higher voltage would have introduced much more error than it would have given back in terms of A/D resolution. Correcting the other readings to the measured reference is just a little math in the firmware, which costs nothing in parts and introduces virtually no additional error. ***************************************************************** Embed Inc, embedded system specialists in Littleton Massachusetts (978) 742-9014, http://www.embedinc.com ____________________________________________

Smart solution :) Note however that if Vdd varies dynamically only +/- 1% between A/D samplings, both the reference and interesting input converted values will vary that much, ending up in +/- 2% additional error after calculation. /Morgan Olin Lathrop 16:29 2004-11-17: >Or you can feed the 2.5V reference into the A/D as another input, run the >A/D reference off the Vdd supply, and do the correction in firmware. This >is what I did in the ProProg. The 2.5V reference from a LM385 goes directly >to an A/D input. The reference voltage is still read with about 1 part in >500 resolution, which is well beyond its accuracy. Multiplying it up to a >higher voltage would have introduced much more error than it would have >given back in terms of A/D resolution. Correcting the other readings to the >measured reference is just a little math in the firmware, which costs >nothing in parts and introduces virtually no additional error. -- Morgan Olsson, Kivik, Sweden ____________________________________________

Morgan Olsson wrote: > Smart solution :) > Note however that if Vdd varies dynamically only +/- 1% between A/D > samplings, both the reference and interesting input converted values > will vary that much, ending up in +/- 2% additional error after > calculation. This is true. In my case I oversampled the signals considerably, then low pass filtered each by 2 or 3 poles (can't remember) before doing the math to get the corrected values. If you need high speed, or as someone else pointed out, have a control loop that can't tolerate some filtering, then this is not a suitable approach. ***************************************************************** Embed Inc, embedded system specialists in Littleton Massachusetts (978) 742-9014, http://www.embedinc.com ____________________________________________